Performance of microstructure measurements using fast-response thermistors attached to a CTD-frame

Turbulence observations have been limited because of difficulty in microstructure measurements. In order to efficiently obtain much more turbulence data down to the ocean floor without spending extra ship-time, we propose a new method, a microstructure profiler attached to a CTD-frame. Since microstructure measurements of velocity shear are sensitive and fragile to noise from instruments, measurements have been performed with free-fall or free-rise instruments whose vibrations to generate noise are minimized. The profiler attached to the CTD-frame can’t suppress vibrations, and we choose fast-response thermistors to measure micro temperature fields, less sensitive to vibrations than velocity fields. However, turbulence data from thermistors have not been common due to their insufficient resolutions. In our study, to overcome this deficiency, correction procedures for thermistor observations are firstly devised by comparing concurrently obtained turbulence data estimated from thermistors and velocity shear probes attached to a free-fall profiler. Turbulence data estimated from thermistors with single pole filter corrections has bias which strongly depends on turbulence intensity. The correction with the form of double pole and 3×10^-3[s] of the time constant is found to be the best match with shear probe data and with much less bias for wide ranges of turbulence intensity. Next, this correction is applied to temperature spectra obtained from thermistors attached to the CTD-frame, and the turbulence intensity is compared with data from the free-fall profiler conducted at the same position within 2 hours. They are similar except when the variation of the fall speed W of the CTD-frame is large. Large (dW/dz)/W tends to violate temperature gradient spectra in wavenumber ranges where Taylor hypothesis are not accomplished, and spectra peaks tend to be obscured. Turbulence intensity estimated from free-fall and CTD-attached thermistors reasonably agree by rejecting spectra with unclear peaks, and spectra with (dW/dz)/W>0.3, where violations of spectra expand to peaks at higher wavenumbers. In future, turbulence observations are expected to expand widely by applying the present method to ship observations and to floats equipped with thermistors.